]> git.ipfire.org Git - thirdparty/suricata.git/log
thirdparty/suricata.git
8 years agodetect: update tests that mix state/stream inspect 2673/head
Victor Julien [Wed, 19 Apr 2017 13:57:31 +0000 (15:57 +0200)] 
detect: update tests that mix state/stream inspect

8 years agounittests: add/improve helpers for stream/flow
Victor Julien [Wed, 19 Apr 2017 13:57:08 +0000 (15:57 +0200)] 
unittests: add/improve helpers for stream/flow

8 years agohttp_header: enable trailer prefilter engines
Victor Julien [Thu, 20 Apr 2017 09:48:06 +0000 (11:48 +0200)] 
http_header: enable trailer prefilter engines

Now that the mpm engines run only for the proper 'progress'
value, the trailing headers need their own engine registration.

8 years agostate: check progress before calling engine
Victor Julien [Tue, 18 Apr 2017 07:38:22 +0000 (09:38 +0200)] 
state: check progress before calling engine

Make sure progress of an inspect engine is available.

8 years agodetect: improve stateful detection
Victor Julien [Tue, 11 Apr 2017 13:24:49 +0000 (15:24 +0200)] 
detect: improve stateful detection

Now that MPM runs when the TX progress is right, stateful detection
operates differently.

Changes:

1. raw stream inspection is now also an inspect engine

   Since this engine doesn't take the transactions into account, it
   could potentially run multiple times on the same data. To avoid
   this, basic result caching is in place.

2. the engines are sorted by progress, but the 'MPM' engine is first
   even if the progress is higher

   If MPM flags a rule to be inspected, the inspect engine for that
   buffer runs first. If this step fails, the rule is no longer
   evaluated. No state is stored.

8 years agodetect: change mask logic
Victor Julien [Tue, 11 Apr 2017 16:15:16 +0000 (18:15 +0200)] 
detect: change mask logic

Previously the MPM/Prefilter engines would suggest the same rule
candidates multiple times.

For example, while processing the request body, the http headers
would be inspected by MPM multiple times.

The mask check was one way to quickly decide which rules could be
skipped.

Now that the MPM engines normally return a rule just once, this
mask check no longer makes sense. If the rule meets the ip/port/
direction based conditions, it needs to be evaluated if the MPM
said so. Even if not all conditions are yet true.

WIP disable mask as it no longer makes sense

WIP redo mask match

8 years agompm: run engines as few times as possible
Victor Julien [Mon, 10 Apr 2017 18:42:25 +0000 (20:42 +0200)] 
mpm: run engines as few times as possible

In various scenarios buffers would be checked my MPM more than
once. This was because the buffers would be inspected for a
certain progress value or higher.

For example, for each packet in a file upload, the engine would
not just rerun the 'http client body' MPM on the new data, it
would also rerun the method, uri, headers, cookie, etc MPMs.

This was obviously inefficent, so this patch changes the logic.

The patch only runs the MPM engines when the progress is exactly
the intended progress. If the progress is beyond the desired
value, it is run once. A tracker is added to the app layer API,
where the completed MPMs are tracked.

Implemented for HTTP, TLS and SSH.

8 years agodetect: register progress in inspect engines
Victor Julien [Mon, 17 Apr 2017 08:59:32 +0000 (10:59 +0200)] 
detect: register progress in inspect engines

Register required progress so we can stop inspecting as soon
as the progress isn't far enough yet.

8 years agodetect-state: don't use casts to uint
Victor Julien [Tue, 18 Apr 2017 07:38:48 +0000 (09:38 +0200)] 
detect-state: don't use casts to uint

8 years agohttp_uri: unittest cleanup
Victor Julien [Thu, 20 Apr 2017 14:15:51 +0000 (16:15 +0200)] 
http_uri: unittest cleanup

8 years agohttp_header: add another trailer test
Victor Julien [Thu, 20 Apr 2017 11:03:42 +0000 (13:03 +0200)] 
http_header: add another trailer test

8 years agohttp_header (trailer) test cleanup
Victor Julien [Thu, 20 Apr 2017 09:30:01 +0000 (11:30 +0200)] 
http_header (trailer) test cleanup

8 years agossh: fix test
Victor Julien [Thu, 20 Apr 2017 14:07:01 +0000 (16:07 +0200)] 
ssh: fix test

8 years agossh: fix banner state setting
Victor Julien [Thu, 20 Apr 2017 14:04:34 +0000 (16:04 +0200)] 
ssh: fix banner state setting

8 years agodetect: more detailed state profiling
Victor Julien [Mon, 17 Apr 2017 07:53:50 +0000 (09:53 +0200)] 
detect: more detailed state profiling

8 years agoapp-layer: set stream-depth after stream init
Victor Julien [Wed, 12 Apr 2017 12:31:25 +0000 (14:31 +0200)] 
app-layer: set stream-depth after stream init

8 years agodetect: clean up stateful detect
Victor Julien [Tue, 11 Apr 2017 07:58:40 +0000 (09:58 +0200)] 
detect: clean up stateful detect

8 years agostream: reduce scope of new ssn func
Victor Julien [Wed, 19 Apr 2017 13:47:09 +0000 (15:47 +0200)] 
stream: reduce scope of new ssn func

8 years agoautotools: add src/tests to extra dist
Victor Julien [Tue, 4 Apr 2017 14:16:17 +0000 (16:16 +0200)] 
autotools: add src/tests to extra dist

8 years agostream: move inline tests
Victor Julien [Wed, 8 Mar 2017 22:28:51 +0000 (23:28 +0100)] 
stream: move inline tests

8 years agostream: list management cleanups
Victor Julien [Wed, 8 Mar 2017 22:20:44 +0000 (23:20 +0100)] 
stream: list management cleanups

8 years agostream: debug improvements
Victor Julien [Wed, 8 Mar 2017 11:52:56 +0000 (12:52 +0100)] 
stream: debug improvements

8 years agostream inspection: add debug counters
Victor Julien [Wed, 8 Mar 2017 11:52:34 +0000 (12:52 +0100)] 
stream inspection: add debug counters

8 years agostream: pack config struct
Victor Julien [Tue, 7 Mar 2017 21:41:23 +0000 (22:41 +0100)] 
stream: pack config struct

8 years agostream: enforce gap earlier in app reassembly
Victor Julien [Fri, 3 Mar 2017 08:54:56 +0000 (09:54 +0100)] 
stream: enforce gap earlier in app reassembly

8 years agostream: don't call app reassembly if disable flag set
Victor Julien [Fri, 3 Mar 2017 08:54:16 +0000 (09:54 +0100)] 
stream: don't call app reassembly if disable flag set

8 years agostream: app-layer micro optimizations
Victor Julien [Fri, 3 Mar 2017 08:53:54 +0000 (09:53 +0100)] 
stream: app-layer micro optimizations

8 years agostream: raw reassembly explicit disable raw handling
Victor Julien [Thu, 2 Mar 2017 23:12:38 +0000 (00:12 +0100)] 
stream: raw reassembly explicit disable raw handling

8 years agostream: mpm inspect micro optimizations
Victor Julien [Thu, 2 Mar 2017 23:12:12 +0000 (00:12 +0100)] 
stream: mpm inspect micro optimizations

8 years agostream: improve --disable-detection GAP handling
Victor Julien [Wed, 1 Mar 2017 15:51:22 +0000 (16:51 +0100)] 
stream: improve --disable-detection GAP handling

8 years agostream: remove unused StreamTcpGetStreamSize function
Victor Julien [Wed, 1 Mar 2017 14:34:14 +0000 (15:34 +0100)] 
stream: remove unused StreamTcpGetStreamSize function

8 years agostream: optimize session pruning
Victor Julien [Wed, 1 Mar 2017 13:50:20 +0000 (14:50 +0100)] 
stream: optimize session pruning

8 years agodoc: update for stream changes
Victor Julien [Wed, 1 Mar 2017 13:04:29 +0000 (14:04 +0100)] 
doc: update for stream changes

8 years agoyaml: sync with new stream engine
Victor Julien [Wed, 1 Mar 2017 10:24:13 +0000 (11:24 +0100)] 
yaml: sync with new stream engine

8 years agostream: validate code
Victor Julien [Tue, 28 Feb 2017 22:13:00 +0000 (23:13 +0100)] 
stream: validate code

8 years agodetect: only do flow dependent cleanup if a flow is present
Victor Julien [Mon, 27 Feb 2017 22:28:46 +0000 (23:28 +0100)] 
detect: only do flow dependent cleanup if a flow is present

8 years agodetect: make SigMatchSignatures void
Victor Julien [Mon, 27 Feb 2017 22:26:37 +0000 (23:26 +0100)] 
detect: make SigMatchSignatures void

None of the callers cared for it's retval, so get rid of it.

8 years agodetect: turn single detect flag into bool
Victor Julien [Mon, 27 Feb 2017 22:14:39 +0000 (23:14 +0100)] 
detect: turn single detect flag into bool

8 years agodetect: remove unused detect flag
Victor Julien [Mon, 27 Feb 2017 22:12:09 +0000 (23:12 +0100)] 
detect: remove unused detect flag

8 years agostream: improve needs reassembly code
Victor Julien [Tue, 28 Feb 2017 11:44:02 +0000 (12:44 +0100)] 
stream: improve needs reassembly code

8 years agostream: more aggressive StreamReassembleRawHasDataReady
Victor Julien [Tue, 28 Feb 2017 10:23:27 +0000 (11:23 +0100)] 
stream: more aggressive StreamReassembleRawHasDataReady

8 years agoapp-layer: change logic of setting 'no reassembly'
Victor Julien [Sun, 26 Feb 2017 13:19:43 +0000 (14:19 +0100)] 
app-layer: change logic of setting 'no reassembly'

Instead of killing all reassembly instantly do things slightly more
gracefully:
1. disable app-layer reassembly immediately
2. flag raw reassembly not to accept new data

This will allow the current data to be inspected still.

After detect as run the raw reassembly will be fully disabled and
thus all reassembly will be as well.

8 years agostream: new depth / disable raw logic
Victor Julien [Sun, 26 Feb 2017 13:06:05 +0000 (14:06 +0100)] 
stream: new depth / disable raw logic

Depth reach sets NOREASSEMBLY after detect.

No new raw sets NORAW after detect.

8 years agostream: allow raw reassembly catch up
Victor Julien [Sat, 25 Feb 2017 12:33:40 +0000 (13:33 +0100)] 
stream: allow raw reassembly catch up

If raw reassembly falls behind, for example because no raw mpm is
active, then we need to sync up to the app progress if that is
available, or to the generic tcp tracking otherwise.

8 years agostream: detect stream GAP also during reassembly
Victor Julien [Sat, 25 Feb 2017 12:33:15 +0000 (13:33 +0100)] 
stream: detect stream GAP also during reassembly

8 years agodebug-validation: add stream checks
Victor Julien [Fri, 24 Feb 2017 20:05:43 +0000 (21:05 +0100)] 
debug-validation: add stream checks

8 years agostream: StreamTcpReassembleRawCheckLimit cleanup
Victor Julien [Thu, 23 Feb 2017 12:09:46 +0000 (13:09 +0100)] 
stream: StreamTcpReassembleRawCheckLimit cleanup

8 years agostream: handle no stream scanning case
Victor Julien [Wed, 22 Feb 2017 10:14:02 +0000 (11:14 +0100)] 
stream: handle no stream scanning case

Now that detect moves the raw progress forward, it's important
to deal with the case where detect don't consider raw inspection.

If no 'stream' rules are active, disable raw. For this the disable
raw flag is now per stream.

8 years agostream: raw content inspection inline mode
Victor Julien [Mon, 20 Feb 2017 10:04:29 +0000 (11:04 +0100)] 
stream: raw content inspection inline mode

Implement the inline mode for raw content inspection. Packets
are leading, and when a packet's payload has been added to the
stream, the packet is inspected in the context of the stream.

Reassembly will return a buffer with the packet data with older
data in front of it and after it, if available.

8 years agoflow/stream: reduce/disable pseudo packet injections
Victor Julien [Sun, 19 Feb 2017 11:32:23 +0000 (12:32 +0100)] 
flow/stream: reduce/disable pseudo packet injections

At flow timeout, we no longer need to first run reassembly in
one dir, then inspection in the other. We can do both in single
packet now.

Disable pseudo packets when receiving stream end packets. Instead
call the app-layer parser in the packet direction for stream end
packets and flow end packets.

These changes in handling of those stream end packets make the
pseudo packets unnecessary.

8 years agostream: set 'trigger raw' per direction
Victor Julien [Sat, 18 Feb 2017 23:54:45 +0000 (00:54 +0100)] 
stream: set 'trigger raw' per direction

8 years agodetect / stream: new 'raw' stream inspection
Victor Julien [Fri, 17 Feb 2017 16:59:43 +0000 (17:59 +0100)] 
detect / stream: new 'raw' stream inspection

Remove the 'StreamMsg' approach from the engine. In this approach the
stream engine would create a list of chunks for inspection by the
detection engine. There were several issues:

1. the messages had a fixed size, so blocks of data bigger than ~4k
   would be cut into multiple messages

2. it lead to lots of data copying and unnecessary memory use

3. the StreamMsgs used a central pool

The Stream engine switched over to the streaming buffer API, which
means that the reassembled data is always available. This made the
StreamMsg approach even clunkier.

The new approach exposes the streaming buffer data to the detection
engine. It has to pay attention to an important issue though: packet
loss. The data may have gaps. The streaming buffer API tracks the
blocks of continuous data.

To access the data for inspection a callback approach is used. The
'StreamReassembleRaw' function is called with a callback and data.
This way it runs the MPM and individual rule inspection code. At
the end of each detection run the stream engine is notified that it
can move forward it's 'progress'.

8 years agostream: constify StreamTcpReassembleRawCheckLimit
Victor Julien [Wed, 8 Mar 2017 12:26:54 +0000 (13:26 +0100)] 
stream: constify StreamTcpReassembleRawCheckLimit

8 years agounittests: fail if TCP memory still in use
Victor Julien [Sat, 7 May 2016 15:24:32 +0000 (17:24 +0200)] 
unittests: fail if TCP memory still in use

abort() so test can be analyzed.

8 years agostream-tcp: implement thread pool for segments
Victor Julien [Fri, 6 May 2016 17:45:30 +0000 (19:45 +0200)] 
stream-tcp: implement thread pool for segments

Config option:

stream:
  reassembly:
    segment-prealloc: 2048

8 years agostream: implement memory handling functions
Victor Julien [Fri, 6 May 2016 15:12:42 +0000 (17:12 +0200)] 
stream: implement memory handling functions

8 years agostream: use static instead of dynamic streaming buffer structure
Victor Julien [Thu, 28 Apr 2016 15:21:28 +0000 (17:21 +0200)] 
stream: use static instead of dynamic streaming buffer structure

8 years agostream: test cleanups and fixes
Victor Julien [Thu, 28 Apr 2016 15:20:11 +0000 (17:20 +0200)] 
stream: test cleanups and fixes

8 years agostream: add insert failure counters
Victor Julien [Wed, 8 Mar 2017 11:50:32 +0000 (12:50 +0100)] 
stream: add insert failure counters

8 years agostream: add stream.reassembly.check-overlap-different-data option
Victor Julien [Thu, 28 Apr 2016 09:48:30 +0000 (11:48 +0200)] 
stream: add stream.reassembly.check-overlap-different-data option

8 years agostream: add tcp.overlap and tcp.overlap_diff_data counters
Victor Julien [Thu, 28 Apr 2016 08:53:49 +0000 (10:53 +0200)] 
stream: add tcp.overlap and tcp.overlap_diff_data counters

8 years agostream: improve no app and no raw case
Victor Julien [Sat, 25 Feb 2017 09:20:51 +0000 (10:20 +0100)] 
stream: improve no app and no raw case

8 years agostream: make raw_progress relative to STREAM_BASE_OFFSET
Victor Julien [Thu, 28 Apr 2016 07:53:24 +0000 (09:53 +0200)] 
stream: make raw_progress relative to STREAM_BASE_OFFSET

8 years agostream: make app_progress relative to STREAM_BASE_OFFSET
Victor Julien [Thu, 28 Apr 2016 06:44:10 +0000 (08:44 +0200)] 
stream: make app_progress relative to STREAM_BASE_OFFSET

8 years agostream: reduce space used for progress tracking
Victor Julien [Wed, 27 Apr 2016 20:13:27 +0000 (22:13 +0200)] 
stream: reduce space used for progress tracking

Instead of the explicit base_seq_offset, use a macro instead. The
macro points to the stream buffer offset. The two were always
in sync.

8 years agostream: small cleanups
Victor Julien [Wed, 27 Apr 2016 19:47:10 +0000 (21:47 +0200)] 
stream: small cleanups

8 years agostream: remove unused zero copy setting
Victor Julien [Wed, 27 Apr 2016 19:37:28 +0000 (21:37 +0200)] 
stream: remove unused zero copy setting

8 years agostream: safety check in overlap handling
Victor Julien [Wed, 8 Mar 2017 22:24:59 +0000 (23:24 +0100)] 
stream: safety check in overlap handling

8 years agotcp: streaming implementation
Victor Julien [Tue, 22 Dec 2015 09:26:04 +0000 (10:26 +0100)] 
tcp: streaming implementation

Make stream engine use the streaming buffer API for it's data storage.

This means that the data is stored in a single reassembled sliding
buffer. The subleties of the reassembly, e.g. overlap handling, are
taken care of at segment insertion.

The TcpSegments now have a StreamingBufferSegment that contains an
offset and a length. Using this the segment data can be retrieved
per segment.

Redo segment insertion. The insertion code is moved to it's own file
and is simplified a lot.

A major difference with the previous implementation is that the segment
list now contains overlapping segments if the traffic is that way.
Previously there could be more and smaller segments in the memory list
than what was seen on the wire.

Due to the matching of in memory segments and on the wire segments,
the overlap with different data detection (potential mots attacks)
is much more accurate.

Raw and App reassembly progress is no longer tracked per segment using
flags, but there is now a progress tracker in the TcpStream for each.

When pruning we make sure we don't slide beyond in-use segments. When
both app-layer and raw inspection are beyond the start of the segment
list, the segments might not be freed even though the data in the
streaming buffer is already gone. This is caused by the 'in-use' status
that the segments can implicitly have. This patch accounts for that
when calculating the 'left_edge' of the streaming window.

Raw reassembly still sets up 'StreamMsg' objects for content
inspection. They are set up based on either the full StreamingBuffer,
or based on the StreamingBufferBlocks if there are gaps in the data.

Reworked 'stream needs work' logic. When a flow times out the flow
engine checks whether a TCP flow still needs work. The
StreamNeedsReassembly function is used to test if a stream still has
unreassembled segments or uninspected stream chunks.

This patch updates the function to consider the app and/or raw
progress. It also cleans the function up and adds more meaningful
debug messages. Finally it makes it non-inline.

Unittests have been overhauled, and partly moved into their own files.

Remove lots of dead code.

8 years agostreaming: remove BUG_ON and other improvements
Victor Julien [Fri, 6 May 2016 17:45:11 +0000 (19:45 +0200)] 
streaming: remove BUG_ON and other improvements

Can be triggered by memory limits.

8 years agostreaming: add blocklist
Victor Julien [Wed, 20 Jan 2016 03:31:22 +0000 (22:31 -0500)] 
streaming: add blocklist

Add list of 'blocks'. This list contains offsets and lengths to
continuous data blocks. This is useful for TCP tracking where we
can have data gaps.

The blocks don't contain any data themselves, instead they contain
lenght and offsets. This way no extra copying is needed.

On inserting new data, existing blocks are expanded instead of
having multiple neighbouring blocks.

8 years agostream-tcp: StreamTcpUTAddPayload unittest helper
Victor Julien [Thu, 17 Mar 2016 09:08:00 +0000 (10:08 +0100)] 
stream-tcp: StreamTcpUTAddPayload unittest helper

8 years agoprofile: account flow-worker tcp-prune step
Victor Julien [Fri, 3 Mar 2017 08:08:23 +0000 (09:08 +0100)] 
profile: account flow-worker tcp-prune step

8 years agodoc: http keywords update
Victor Julien [Wed, 5 Apr 2017 08:33:23 +0000 (10:33 +0200)] 
doc: http keywords update

8 years agodetect: implement http_content_len sticky buffer
Victor Julien [Sat, 1 Apr 2017 21:58:51 +0000 (23:58 +0200)] 
detect: implement http_content_len sticky buffer

This implements inspection of the Content-Length buffer as a content
sticky buffer.

8 years agodetect: implement http_content_type sticky buffer
Victor Julien [Sat, 1 Apr 2017 21:28:12 +0000 (23:28 +0200)] 
detect: implement http_content_type sticky buffer

8 years agodetect: implement http_accept_enc sticky buffer
Victor Julien [Sat, 1 Apr 2017 10:38:46 +0000 (12:38 +0200)] 
detect: implement http_accept_enc sticky buffer

Inspects Accept-Encoding header.

8 years agodetect: implement http_accept_lang sticky buffer
Victor Julien [Sat, 1 Apr 2017 10:33:49 +0000 (12:33 +0200)] 
detect: implement http_accept_lang sticky buffer

Inspects Accept-Language header

8 years agodetect: implement http_connection sticky buffer
Victor Julien [Sat, 1 Apr 2017 10:23:05 +0000 (12:23 +0200)] 
detect: implement http_connection sticky buffer

8 years agodetect: implement http referer sticky buffer
Victor Julien [Sat, 1 Apr 2017 10:13:17 +0000 (12:13 +0200)] 
detect: implement http referer sticky buffer

8 years agodetect: http_accept sticky buffer + common code
Victor Julien [Sat, 1 Apr 2017 09:49:20 +0000 (11:49 +0200)] 
detect: http_accept sticky buffer + common code

Implement common code to easily add more per HTTP header detection
keywords.

Implement http_accept sticky buffer. It operates on the HTTP Accept
header.

8 years agotls: logging for session resumption 2672/head
Ray Ruvinskiy [Thu, 24 Nov 2016 17:16:09 +0000 (12:16 -0500)] 
tls: logging for session resumption

We assume session resumption has occurred if the Client Hello message
included a session id, we have not seen the server certificate, but
we have seen a Change Cipher Spec message from the server.

Previously, these transactions were not logged at all because the
server cert was never seen.

Ticket: https://redmine.openinfosecfoundation.org/issues/1969

8 years agodevice: fix warning about NULL device
Ray Ruvinskiy [Wed, 19 Apr 2017 15:27:26 +0000 (11:27 -0400)] 
device: fix warning about NULL device

Fix '[ERRCODE: SC_ERR_INVALID_VALUE(130)] - Name of device should not be null'
warning on start-up with pfring.

Ticket: https://redmine.openinfosecfoundation.org/issues/2097

8 years agoredis: use SCCalloc to reduce risk of unitialized vars
Victor Julien [Thu, 20 Apr 2017 07:32:01 +0000 (09:32 +0200)] 
redis: use SCCalloc to reduce risk of unitialized vars

8 years agoeve: async mode for redis output
fooinha [Thu, 23 Feb 2017 22:42:05 +0000 (22:42 +0000)] 
eve: async mode for redis output

eve: detects libevent for async redis at configure
eve: moves redis output code to new file - util-log-redis.{c,h}
eve: redis ECHO and QUIT commands for async mode
eve: redis output defaults if conf is missing

8 years agodoc: async mode for redis eve output
fooinha [Thu, 23 Feb 2017 22:42:51 +0000 (22:42 +0000)] 
doc: async mode for redis eve output

async: true ## if redis replies are read asynchronously

8 years agoqa: add --no-random commandline option 2666/head
Victor Julien [Sat, 15 Apr 2017 16:58:31 +0000 (18:58 +0200)] 
qa: add --no-random commandline option

8 years agoqa: add rand/rand_r to banned functions
Victor Julien [Sat, 15 Apr 2017 21:34:18 +0000 (23:34 +0200)] 
qa: add rand/rand_r to banned functions

8 years agorandom: convert stream and htp to new call
Victor Julien [Sat, 15 Apr 2017 21:29:48 +0000 (23:29 +0200)] 
random: convert stream and htp to new call

8 years agorandom: improve random logic
Victor Julien [Sat, 15 Apr 2017 21:20:48 +0000 (23:20 +0200)] 
random: improve random logic

Improve random logic for hash tables.

Implement Windows random API if it is available.

8 years agomingw: don't use uint type as mingw doesn't have it
Victor Julien [Thu, 13 Apr 2017 14:55:34 +0000 (16:55 +0200)] 
mingw: don't use uint type as mingw doesn't have it

8 years agocommon: improve byte order and wordsize detection 2661/head
Victor Julien [Mon, 10 Apr 2017 07:39:23 +0000 (09:39 +0200)] 
common: improve byte order and wordsize detection

8 years agocore dumps: check for sys/resource.h
Victor Julien [Mon, 10 Apr 2017 07:38:59 +0000 (09:38 +0200)] 
core dumps: check for sys/resource.h

8 years agocleanup: get rid of %llu format specifiers
Victor Julien [Sun, 9 Apr 2017 19:04:14 +0000 (21:04 +0200)] 
cleanup: get rid of %llu format specifiers

Use more explicit types instead.

8 years agodoc: Napatech docs improvement
psanders240 [Thu, 16 Mar 2017 19:41:16 +0000 (15:41 -0400)] 
doc: Napatech docs improvement

Fix errors and simplify filters.

8 years agotemplate script: typo in app-layer setup script
Jason Ish [Mon, 10 Apr 2017 04:12:12 +0000 (22:12 -0600)] 
template script: typo in app-layer setup script

Check for ed was failing, as it was actually looking for edx.

8 years agopcre: on ppc64 disable only for specific versions
Victor Julien [Thu, 13 Apr 2017 07:58:36 +0000 (09:58 +0200)] 
pcre: on ppc64 disable only for specific versions

Disable jit only for libpcre 8.39 and 8.40 as those were the buggy
versions.

Thanks to Zoltán Herczeg.

8 years agodoc: expand on bpf
Victor Julien [Thu, 13 Apr 2017 07:25:11 +0000 (09:25 +0200)] 
doc: expand on bpf

8 years agodetect: don't consider an empty rule file an error 2655/head
Jason Ish [Wed, 15 Mar 2017 16:52:57 +0000 (10:52 -0600)] 
detect: don't consider an empty rule file an error

8 years agologging: only do non-blocking writes if live
Jason Ish [Wed, 5 Apr 2017 04:44:21 +0000 (22:44 -0600)] 
logging: only do non-blocking writes if live

If running against a pcap there is no reason to drop events,
a blocking socket is fine here. So only do non-blocking writes
when running off a live device.